/**
* @covers DateStringNormalizerFilter
*/
public function testExecute()
{
$filter = new DateStringNormalizerFilter();
self::assertEquals('2003', $filter->execute($dateString = ' 2003 '));
self::assertEquals('2003-07', $filter->execute($dateString = ' 2003 Jul '));
self::assertEquals('2003-07-05', $filter->execute($dateString = ' 2003 Jul 5 '));
self::assertEquals('2003', $filter->execute($dateString = ' 2003 5 '));
self::assertNull($filter->execute($dateString = 'unparsable string'));
}
/**
* Custom implementation of Form::validate() that validates
* meta-data form data and injects it into the internal citation
* object.
*
* NB: The configuration of the internal citation object
* would normally be done in readInputData(). Validation and
* injection can easily be done in one step. It therefore avoids
* code duplication and improves performance to do both here.
*/
function validate()
{
// Make sure that this method is not called twice which
// would corrupt internal state.
assert(empty($this->_metadataDescriptions));
parent::validate();
// Validate form data and inject it into
// the associated citation object.
$citation =& $this->getCitation();
$citation->setRawCitation($this->getData('rawCitation'));
if ($this->getData('citationApproved') == 'citationApproved') {
// Editor's shortcut to the approved state, e.g. for manually edited citations.
$citation->setCitationState(CITATION_APPROVED);
} elseif (in_array($this->getData('citationState'), Citation::_getSupportedCitationStates())) {
// Reset citation state if necessary
if ($this->getData('citationState') == CITATION_APPROVED) {
$this->setData('citationState', CITATION_LOOKED_UP);
}
$citation->setCitationState($this->getData('citationState'));
}
// Extract data from citation form fields and inject it into the citation
import('lib.pkp.classes.metadata.MetadataDescription');
$metadataSchemas = $citation->getSupportedMetadataSchemas();
foreach ($metadataSchemas as $metadataSchema) {
/* @var $metadataSchema MetadataSchema */
// Instantiate a meta-data description for the given schema
$metadataDescription = new MetadataDescription($metadataSchema->getClassName(), ASSOC_TYPE_CITATION);
// Set the meta-data statements
foreach ($metadataSchema->getProperties() as $propertyName => $property) {
$fieldName = $metadataSchema->getNamespacedPropertyId($propertyName);
$fieldValue = trim($this->getData($fieldName));
if (empty($fieldValue)) {
// Delete empty statements so that previously set
// statements (if any) will be deleted.
$metadataDescription->removeStatement($propertyName);
if ($property->getMandatory()) {
// A mandatory field is missing - add a validation error.
$this->addError($fieldName, __($property->getValidationMessage()));
$this->addErrorField($fieldName);
}
} else {
// Try to convert the field value to (a) strongly
// typed object(s) if applicable. Start with the most
// specific allowed type so that we always get the
// most strongly typed result possible.
$allowedTypes = $property->getAllowedTypes();
switch (true) {
case isset($allowedTypes[METADATA_PROPERTY_TYPE_VOCABULARY]) && is_numeric($fieldValue):
case isset($allowedTypes[METADATA_PROPERTY_TYPE_INTEGER]) && is_numeric($fieldValue):
$typedFieldValues = array((int) $fieldValue);
break;
case isset($allowedTypes[METADATA_PROPERTY_TYPE_DATE]):
import('lib.pkp.classes.metadata.DateStringNormalizerFilter');
$dateStringFilter = new DateStringNormalizerFilter();
assert($dateStringFilter->supportsAsInput($fieldValue));
$typedFieldValues = array($dateStringFilter->execute($fieldValue));
break;
case isset($allowedTypes[METADATA_PROPERTY_TYPE_COMPOSITE]):
// We currently only support name composites
$allowedAssocIds = $allowedTypes[METADATA_PROPERTY_TYPE_COMPOSITE];
if (in_array(ASSOC_TYPE_AUTHOR, $allowedAssocIds)) {
$assocType = ASSOC_TYPE_AUTHOR;
} elseif (in_array(ASSOC_TYPE_EDITOR, $allowedAssocIds)) {
$assocType = ASSOC_TYPE_EDITOR;
} else {
assert(false);
}
// Try to transform the field to a name composite.
import('lib.pkp.plugins.metadata.nlm30.filter.PersonStringNlm30NameSchemaFilter');
$personStringFilter = new PersonStringNlm30NameSchemaFilter($assocType, PERSON_STRING_FILTER_MULTIPLE);
assert($personStringFilter->supportsAsInput($fieldValue));
$typedFieldValues =& $personStringFilter->execute($fieldValue);
break;
default:
$typedFieldValues = array($fieldValue);
}
// Inject data into the meta-data description and thereby
// implicitly validate the field value.
foreach ($typedFieldValues as $typedFieldValue) {
if (!$metadataDescription->addStatement($propertyName, $typedFieldValue)) {
// Add form field error
$this->addError($fieldName, __($property->getValidationMessage()));
$this->addErrorField($fieldName);
}
unset($typedFieldValue);
}
unset($typedFieldValues);
}
}
// Inject the meta-data into the citation.
$citation->injectMetadata($metadataDescription);
// Save the meta-data description for later usage.
$this->_metadataDescriptions[] =& $metadataDescription;
unset($metadataDescription);
}
return $this->isValid();
}
/**
* Fills the given citation object with
* meta-data retrieved from PubMed.
* @param $pmid string
* @return MetadataDescription
*/
function &_lookup($pmid)
{
$nullVar = null;
// Use eFetch to get XML metadata for the given PMID
$lookupParams = array('db' => 'pubmed', 'mode' => 'xml', 'tool' => 'pkp-wal', 'id' => $pmid);
if (!is_null($this->getEmail())) {
$lookupParams['email'] = $this->getEmail();
}
// Call the eFetch URL and get an XML result
if (is_null($resultDOM = $this->callWebService(PUBMED_WEBSERVICE_EFETCH, $lookupParams))) {
return $nullVar;
}
$articleTitleNodes =& $resultDOM->getElementsByTagName("ArticleTitle");
$articleTitleFirstNode =& $articleTitleNodes->item(0);
$medlineTaNodes =& $resultDOM->getElementsByTagName("MedlineTA");
$medlineTaFirstNode =& $medlineTaNodes->item(0);
$metadata = array('pub-id[@pub-id-type="pmid"]' => $pmid, 'article-title' => $articleTitleFirstNode->textContent, 'source' => $medlineTaFirstNode->textContent);
$volumeNodes =& $resultDOM->getElementsByTagName("Volume");
$issueNodes =& $resultDOM->getElementsByTagName("Issue");
if ($volumeNodes->length > 0) {
$volumeFirstNode =& $volumeNodes->item(0);
}
$metadata['volume'] = $volumeFirstNode->textContent;
if ($issueNodes->length > 0) {
$issueFirstNode =& $issueNodes->item(0);
}
$metadata['issue'] = $issueFirstNode->textContent;
// Get list of author full names
foreach ($resultDOM->getElementsByTagName("Author") as $authorNode) {
if (!isset($metadata['person-group[@person-group-type="author"]'])) {
$metadata['person-group[@person-group-type="author"]'] = array();
}
// Instantiate an NLM name description
$authorDescription = new MetadataDescription('lib.pkp.plugins.metadata.nlm30.schema.Nlm30NameSchema', ASSOC_TYPE_AUTHOR);
// Surname
$lastNameNodes =& $authorNode->getElementsByTagName("LastName");
$lastNameFirstNode =& $lastNameNodes->item(0);
$authorDescription->addStatement('surname', $lastNameFirstNode->textContent);
// Given names
$givenNamesString = '';
$firstNameNodes =& $authorNode->getElementsByTagName("FirstName");
if ($firstNameNodes->length > 0) {
$firstNameFirstNode =& $firstNameNodes->item(0);
$givenNamesString = $firstNameFirstNode->textContent;
} else {
$foreNameNodes =& $authorNode->getElementsByTagName("ForeName");
if ($foreNameNodes->length > 0) {
$foreNameFirstNode =& $foreNameNodes->item(0);
$givenNamesString = $foreNameFirstNode->textContent;
}
}
if (!empty($givenNamesString)) {
foreach (explode(' ', $givenNamesString) as $givenName) {
$authorDescription->addStatement('given-names', String::trimPunctuation($givenName));
}
}
// Suffix
$suffixNodes =& $authorNode->getElementsByTagName("Suffix");
if ($suffixNodes->length > 0) {
$suffixFirstNode =& $suffixNodes->item(0);
$authorDescription->addStatement('suffix', $suffixFirstNode->textContent);
}
// Include collective names
// FIXME: This corresponds to an NLM-citation <collab> tag and should be part of the Metadata implementation
/*if ($resultDOM->getElementsByTagName("CollectiveName")->length > 0 && $authorNode->getElementsByTagName("CollectiveName")->item(0)->textContent != '') {
}*/
$metadata['person-group[@person-group-type="author"]'][] =& $authorDescription;
unset($authorDescription);
}
// Extract pagination
$medlinePgnNodes =& $resultDOM->getElementsByTagName("MedlinePgn");
$medlinePgnFirstNode =& $medlinePgnNodes->item(0);
if (String::regexp_match_get("/^[:p\\.\\s]*(?P<fpage>[Ee]?\\d+)(-(?P<lpage>\\d+))?/", $medlinePgnFirstNode->textContent, $pages)) {
$fPage = (int) $pages['fpage'];
$metadata['fpage'] = $fPage;
if (!empty($pages['lpage'])) {
$lPage = (int) $pages['lpage'];
// Deal with shortcuts like '382-7'
if ($lPage < $fPage) {
$lPage = (int) (String::substr($pages['fpage'], 0, -String::strlen($pages['lpage'])) . $pages['lpage']);
}
$metadata['lpage'] = $lPage;
}
}
// Get publication date (can be in several places in PubMed).
$dateNode = null;
$articleDateNodes =& $resultDOM->getElementsByTagName("ArticleDate");
if ($articleDateNodes->length > 0) {
$dateNode =& $articleDateNodes->item(0);
} else {
$pubDateNodes =& $resultDOM->getElementsByTagName("PubDate");
if ($pubDateNodes->length > 0) {
$dateNode =& $pubDateNodes->item(0);
}
}
// Retrieve the data parts and assemble date.
if (!is_null($dateNode)) {
$publicationDate = '';
$requiresNormalization = false;
foreach (array('Year' => 4, 'Month' => 2, 'Day' => 2) as $dateElement => $padding) {
$dateElementNodes =& $dateNode->getElementsByTagName($dateElement);
if ($dateElementNodes->length > 0) {
if (!empty($publicationDate)) {
$publicationDate .= '-';
}
$dateElementFirstNode =& $dateElementNodes->item(0);
$datePart = str_pad($dateElementFirstNode->textContent, $padding, '0', STR_PAD_LEFT);
if (!is_numeric($datePart)) {
$requiresNormalization = true;
}
$publicationDate .= $datePart;
} else {
break;
}
}
// Normalize the date to NLM standard if necessary.
if ($requiresNormalization) {
$dateFilter = new DateStringNormalizerFilter();
$publicationDate = $dateFilter->execute($publicationDate);
}
if (!empty($publicationDate)) {
$metadata['date'] = $publicationDate;
}
}
// Get publication type
$publicationTypeNodes =& $resultDOM->getElementsByTagName("PublicationType");
if ($publicationTypeNodes->length > 0) {
foreach ($publicationTypeNodes as $publicationType) {
// The vast majority of items on PubMed are articles so catch these...
if (String::strpos(String::strtolower($publicationType->textContent), 'article') !== false) {
$metadata['[@publication-type]'] = NLM30_PUBLICATION_TYPE_JOURNAL;
break;
}
}
}
// Get DOI if it exists
$articleIdNodes =& $resultDOM->getElementsByTagName("ArticleId");
foreach ($articleIdNodes as $idNode) {
if ($idNode->getAttribute('IdType') == 'doi') {
$metadata['pub-id[@pub-id-type="doi"]'] = $idNode->textContent;
}
}
// Use eLink utility to find fulltext links
$lookupParams = array('dbfrom' => 'pubmed', 'cmd' => 'llinks', 'tool' => 'pkp-wal', 'id' => $pmid);
if (!is_null($resultDOM = $this->callWebService(PUBMED_WEBSERVICE_ELINK, $lookupParams))) {
// Get a list of possible links
foreach ($resultDOM->getElementsByTagName("ObjUrl") as $linkOut) {
$attributes = '';
foreach ($linkOut->getElementsByTagName("Attribute") as $attribute) {
$attributes .= String::strtolower($attribute->textContent) . ' / ';
}
// Only add links to open access resources
if (String::strpos($attributes, "subscription") === false && String::strpos($attributes, "membership") === false && String::strpos($attributes, "fee") === false && $attributes != "") {
$urlNodes =& $linkOut->getElementsByTagName("Url");
$urlFirstNode =& $urlNodes->item(0);
$links[] = $urlFirstNode->textContent;
}
}
// Take the first link if we have any left (presumably pubmed returns them in preferential order)
if (isset($links[0])) {
$metadata['uri'] = $links[0];
}
}
return $this->getNlm30CitationDescriptionFromMetadataArray($metadata);
}
/**
* Post processes an NLM meta-data array
* @param $preliminaryNlm30Array array
* @return array
*/
function &postProcessMetadataArray(&$preliminaryNlm30Array)
{
// Clean array
$preliminaryNlm30Array = arrayClean($preliminaryNlm30Array);
// Trim punctuation
$preliminaryNlm30Array =& $this->_recursivelyTrimPunctuation($preliminaryNlm30Array);
// Parse (=filter) author/editor strings into NLM name descriptions
foreach (array('author' => ASSOC_TYPE_AUTHOR, 'editor' => ASSOC_TYPE_EDITOR) as $personType => $personAssocType) {
if (isset($preliminaryNlm30Array[$personType])) {
// Get the author/editor strings from the result
$personStrings = $preliminaryNlm30Array[$personType];
unset($preliminaryNlm30Array[$personType]);
// Parse the author/editor strings into NLM name descriptions
// Interpret a scalar as a textual authors list
if (is_scalar($personStrings)) {
$personStringFilter = new PersonStringNlm30NameSchemaFilter($personAssocType, PERSON_STRING_FILTER_MULTIPLE);
$persons =& $personStringFilter->execute($personStrings);
} else {
$personStringFilter = new PersonStringNlm30NameSchemaFilter($personAssocType, PERSON_STRING_FILTER_SINGLE);
$persons =& array_map(array($personStringFilter, 'execute'), $personStrings);
}
$preliminaryNlm30Array['person-group[@person-group-type="' . $personType . '"]'] = $persons;
unset($persons);
}
}
// Join comments
if (isset($preliminaryNlm30Array['comment']) && is_array($preliminaryNlm30Array['comment'])) {
// Implode comments from the result into a single string
// as required by the NLM citation schema.
$preliminaryNlm30Array['comment'] = implode("\n", $preliminaryNlm30Array['comment']);
}
// Normalize date strings
foreach (array('date', 'conf-date', 'access-date') as $dateProperty) {
if (isset($preliminaryNlm30Array[$dateProperty])) {
$dateFilter = new DateStringNormalizerFilter();
$preliminaryNlm30Array[$dateProperty] = $dateFilter->execute($preliminaryNlm30Array[$dateProperty]);
}
}
// Cast strings to integers where necessary
foreach (array('fpage', 'lpage', 'size') as $integerProperty) {
if (isset($preliminaryNlm30Array[$integerProperty]) && is_numeric($preliminaryNlm30Array[$integerProperty])) {
$preliminaryNlm30Array[$integerProperty] = (int) $preliminaryNlm30Array[$integerProperty];
}
}
// Rename elements that are stored in attributes in NLM citation
$elementToAttributeMap = array('access-date' => 'date-in-citation[@content-type="access-date"]', 'issn-ppub' => 'issn[@pub-type="ppub"]', 'issn-epub' => 'issn[@pub-type="epub"]', 'pub-id-doi' => 'pub-id[@pub-id-type="doi"]', 'pub-id-publisher-id' => 'pub-id[@pub-id-type="publisher-id"]', 'pub-id-coden' => 'pub-id[@pub-id-type="coden"]', 'pub-id-sici' => 'pub-id[@pub-id-type="sici"]', 'pub-id-pmid' => 'pub-id[@pub-id-type="pmid"]', 'publication-type' => '[@publication-type]');
foreach ($elementToAttributeMap as $elementName => $nlm30PropertyName) {
if (isset($preliminaryNlm30Array[$elementName])) {
$preliminaryNlm30Array[$nlm30PropertyName] = $preliminaryNlm30Array[$elementName];
unset($preliminaryNlm30Array[$elementName]);
}
}
// Guess a publication type if none has been set by the
// citation service.
$this->_guessPublicationType($preliminaryNlm30Array);
// Some services return the title as article-title although
// the publication type is a book.
if (isset($preliminaryNlm30Array['[@publication-type]']) && $preliminaryNlm30Array['[@publication-type]'] == 'book') {
if (isset($preliminaryNlm30Array['article-title']) && !isset($preliminaryNlm30Array['source'])) {
$preliminaryNlm30Array['source'] = $preliminaryNlm30Array['article-title'];
unset($preliminaryNlm30Array['article-title']);
}
}
return $preliminaryNlm30Array;
}
/**
* Save citation
*/
function execute()
{
$citation =& $this->getCitation();
$citation->setEditedCitation($this->getData('editedCitation'));
if (in_array($this->getData('citationState'), Citation::_getSupportedCitationStates())) {
$citation->setCitationState($this->getData('citationState'));
}
// Extract data from citation form fields and inject it into the citation
$metadataAdapters = $citation->getSupportedMetadataAdapters();
foreach ($metadataAdapters as $metadataAdapter) {
// Instantiate a meta-data description for the given schema
$metadataSchema =& $metadataAdapter->getMetadataSchema();
import('metadata.MetadataDescription');
$metadataDescription = new MetadataDescription($metadataSchema, ASSOC_TYPE_CITATION);
// Set the meta-data statements
$metadataSchemaNamespace = $metadataSchema->getNamespace();
foreach ($metadataSchema->getProperties() as $propertyName => $property) {
$fieldName = $metadataSchema->getNamespacedPropertyId($propertyName);
$fieldValue = trim($this->getData($fieldName));
if (empty($fieldValue)) {
$metadataDescription->removeStatement($propertyName);
} else {
$foundValidType = false;
foreach ($property->getTypes() as $type) {
// Some property types need to be converted first
switch ($type) {
// We currently only support name composites
case array(METADATA_PROPERTY_TYPE_COMPOSITE => ASSOC_TYPE_AUTHOR):
case array(METADATA_PROPERTY_TYPE_COMPOSITE => ASSOC_TYPE_EDITOR):
import('metadata.nlm.PersonStringNlmNameSchemaFilter');
$personStringFilter = new PersonStringNlmNameSchemaFilter($type[METADATA_PROPERTY_TYPE_COMPOSITE], PERSON_STRING_FILTER_MULTIPLE);
assert($personStringFilter->supportsAsInput($fieldValue));
$fieldValue =& $personStringFilter->execute($fieldValue);
$foundValidType = true;
break;
case METADATA_PROPERTY_TYPE_INTEGER:
$fieldValue = array((int) $fieldValue);
$foundValidType = true;
break;
case METADATA_PROPERTY_TYPE_DATE:
import('metadata.DateStringNormalizerFilter');
$dateStringFilter = new DateStringNormalizerFilter();
assert($dateStringFilter->supportsAsInput($fieldValue));
$fieldValue = array($dateStringFilter->execute($fieldValue));
$foundValidType = true;
break;
default:
if ($property->isValid($fieldValue)) {
$fieldValue = array($fieldValue);
$foundValidType = true;
break;
}
}
// Break the outer loop once we found a valid
// interpretation for our form field.
if ($foundValidType) {
break;
}
}
foreach ($fieldValue as $fieldValueStatement) {
$metadataDescription->addStatement($propertyName, $fieldValueStatement);
unset($fieldValueStatement);
}
}
}
// Inject the meta-data into the citation
$citation->injectMetadata($metadataDescription, true);
}
// Persist citation
$citationDAO =& DAORegistry::getDAO('CitationDAO');
if (is_numeric($citation->getId())) {
$citationDAO->updateCitation($citation);
} else {
$citationDAO->insertCitation($citation);
}
return true;
}
